There are a few clarifications in this excellent article that appeared in
the publication of the American Banking Association.  As of October 1,
there were more than 1,200 talking ATM's in the United States.  A number
of these machines can provide a balance inquiry for the end user.  Some
banks, such as Bank of America, use recorded voice files but the voice is
synthetic speech rather than human speech.

Kelly

American Banker
January 4, 2002

HOW ATM OPERATORS CAN PICK THE BEST 'VOICE' FOR MACHINES

At some point over the next 12 to 18 months the Justice Department will
publish additional Americans with Disabilities Act regulations mandating
new automated teller machine capabilities to meet the needs of the
visually impaired.

One of the most significant of the new requirements proposed, and one
that is certain to be implemented, will specify that "Machines shall
provide visual and audible instructions for operation."

While a number of banking companies have already begun to install
talking ATMs, there are probably still fewer than 1,000 in operation in
the United States. Most of them cannot speak the full slate of
transactions the ATM is designed to perform, nor can they speak balance
and receipt information. Such limitations will virtually guarantee that
these ATMs will not satisfy the coming mandate. There are currently two
technologies capable of satisfying the proposed ADA requirements. The
first, used in limited form by the talking ATMs already installed, uses
the WAV file technology familiar to many home PC users. The second, just
now coming to market, is text-to-speech synthesis technology (SST).
Today financial institutions face a critical choice in determining which
of these technologies to embrace.

Essentially, WAV files are digitized recordings of real human voices.
They must be recorded in advance by a human performer and are relatively
inflexible. To speak a simple amount, an ATM using this technology must
string together individual files for each component of that amount.

For example, to inform a visually impaired customer that his or her
balance is $ 1,342.22, the ATM must first invoke a WAV file for "one,"
then for "thousand," then for "three," then for "hundred," and so on
down to "cents."

While automated devices can perform this operation very quickly, the
reliance on WAV file technology requires that the ATM or its host
maintain a complete set of recordings for each possible number, each
transaction option, and each transaction screen shown on the terminal.
If the operators wish to present languages other than English, then they
must provide a complete, alternate set of files in the desired language.

The most serious limitation of this technology is that the machine
cannot speak anything that the ATM operator has not previously
anticipated and recorded. It cannot, for instance, accommodate dynamic
information such as customer-generated account names. If a customer
wants to label his savings account "Jamaica Vacation Trust Fund," then
the ATM operator must either suppress the option or schedule a new
recording session.

SST-enabled ATMs, by contrast, can turn any text that an embedded PC or
host computer can generate into audible speech, either through software
routines or dedicated chips incorporated into the machine's hardware.
They can thus speak all the transactional and balance information that
is presented visually on the screen or on printed receipts and coupons,
as well as any dynamic information that the customer or bank may
generate.

Because the technology does not require recordings of human voices, it
eliminates the costs of "talent" and professional recording sessions,
and the machines can operate in multiple languages.

The sole disadvantage of SST technology as opposed to WAV file
technology is that the SST-generated voice sounds robotic. However, this
does not affect the function of the machines, and market research
indicates that the clarity of an ATM's voice is vastly more important to
visually impaired customers than human-like timbre.

The advantages of SST technology will become most readily apparent as
ATM operators seek to upgrade their machines by adding such capabilities
as choice of language, Web-enabled sales and marketing, or on-screen
advertising and interactive marketing.

With WAV file technology, ATM operators would be obliged to record new
messages for every new option and feature, and then upload these files
to all their individual machines. More flexible SST-based devices will
be far easier and more economical to upgrade and maintain.

The deployment of talking ATMs can help banks generate more business
with many other customer segments besides the blind. Voice technology
will encourage ATM use among customers with dyslexia and related forms
of learning disability, the illiterate and barely literate (a
low-profile but surprisingly numerous group), and visitors or immigrants
who may speak English but do not read it.

The new technology might even attract the elderly, a market segment
known for preferring tellers to low-cost ATMs.

While the provisions of the coming ADA regulations are still under
development, their technological implications are clear. Whatever the
final details, the coming regulations will mandate significant new voice
capabilities for ATMs. Smart bankers will take steps today to turn these
requirements to their advantage.


VICUG-L is the Visually Impaired Computer User Group List.
To join or leave the list, send a message to
[log in to unmask]  In the body of the message, simply type
"subscribe vicug-l" or "unsubscribe vicug-l" without the quotations.
 VICUG-L is archived on the World Wide Web at
http://maelstrom.stjohns.edu/archives/vicug-l.html